Automatic 3D graph cuts for brain cortex segmentation in patients with focal cortical dysplasia

Area of Science:

• Medical Imaging
• Neuroscience
• Computer Vision

Background:

• Focal cortical dysplasia (FCD) is a leading cause of intractable epilepsy, often diagnosed via subtle MRI abnormalities.
• Accurate brain cortex segmentation is a critical prerequisite for reliable FCD lesion detection using MRI.
• Existing segmentation methods may lack the accuracy needed for subtle FCD identification.

Purpose of the Study:

• To develop and validate an improved graph cuts algorithm for accurate 3D brain MRI cortex segmentation.
• To enhance the detection of focal cortical dysplasia (FCD) lesions by improving the preprocessing segmentation step.
• To integrate intensity and boundary information using a Markov random field-based energy function for robust segmentation.

Main Methods:

• Proposed an enhanced graph cuts algorithm utilizing a three-label graph cut approach.
• Integrated a Markov random field-based energy function to incorporate intensity and boundary information.
• Validated the method on simulated MR brain images with varying noise levels and real patient data with FCD.

Main Results:

• The proposed method demonstrated high accuracy and robustness to noise in quantitative segmentation results.
• Achieved superior performance compared to other popular brain MRI segmentation techniques.
• Qualitative validation confirmed effective segmentation of FCD lesions, highlighting cortical thickening and deformation.

Conclusions:

• The improved graph cuts algorithm provides accurate and robust brain cortex segmentation for FCD detection.
• This technique offers significant potential for improving the diagnosis of epilepsy and other neurological conditions.
• The method shows promise for broader clinical applications requiring precise brain MRI segmentation.